Drivers with a high level of trait anger feel more intensity of anger on road, contributing to more risky driving behavior and further increasing the probability of collisions. It seems that trait anger directly correlates with risky driving behavior, but how it works in detail remains unknown and previous research indicated executive function and hazard cognition may play a mediation role in it. Our research aims to explore the relationship among these variables and test if there is a multiple mediation model. We sampled 302 valid participants and used online questionnaires, containing trait anger scale (TAS), executive function index (EFI), hazard cognition scale (HCS; representing attitudes towards risky driving behavior), driver behavior questionnaire (DBQ), and self-reported traffic violations (e.g., accidents, penalty points, fines). Hierarchical multiple linear regression of DBQ results show trait anger is a medium but statistically significant predictor of risky driving behavior and drivers’ attitude towards risky situations can significantly predict risky driving behavior at medium effect. But risky driving behavior cannot be predicted by executive function. Interestingly, opposing to prior research, zero-inflated Poisson regression analysis of self-reported traffic violations suggests trait anger negatively predicts accidents and fines in the zero-inflation model, and hazard cognition negatively predicts penalty points. Notably, the executive function negatively predicts penalty points and fines in the count model, which confirms our hypothetical direction. They all represent a small effect size in this nonlinear regression model. Path analysis suggested that trait anger influences risky driving behavior through executive function, and hazard cognition both separately and jointly. This study provides a theoretical framework for the transaction model of aggressive driving behavior and offers some possible interventions toward the effect of trait anger on risky driving behavior.